Commands will time out if they are unable to complete after a period of time.

If necessary, you can abort a pending command. For example, this may become necessary because of a hardware or network failure where a host running a role instance becomes disconnected from the Cloudera Manager Server host. In this case, the Cloudera Manager Server will be unable to connect to the Cloudera Manager Agent on that disconnected host to start or stop the role instance which will prevent the corresponding service from starting or stopping. To work around this, you can abort the command to start or stop the role instance on the disconnected host, and then you can start or stop the service again.

To abort any pending command:

You can click the indicator () with the blue badge, which shows the number of commands that are currently running in your cluster (if any). This indicator is positioned just to the left of the Support link at the right hand side of the navigation bar. Unlike the Commands tab for a role or service, this indicator includes all commands running for all services or roles in the cluster. In the Running Commands window, click Abort to abort the pending command. For more information, see Viewing Running and Recent Commands.

To abort a pending command for a service or role:

1. Navigate to the Service > Instances tab for the service where the role instance you want to stop is located. For example, navigate to the HDFS Service > Instances tab if you want to abort a pending command for a DataNode.
2. In the list of instances, click the link for role instance where the command is running (for example, the instance that is located on the disconnected host).
3. Go to the Commands tab.
4. Find the command in the list of Running Commands and click Abort Command to abort the running command.